MARP, Inc.
Physicist Role in ACR MRI Accreditation
May 2008
Carl R. Keener, Ph.D., DABMP, DABR
keener@MARPinc.com
MARP
Medical & Radiation Physics, Inc.
Department of Radiology
University of Texas Health Science Center at San Antonio
ACR accreditation process & phantom
MARP, Inc.
United Healthcare will require accreditation for
reimbursement by 3 quarter of 2008
MRAP (ACR)
ICAMRL (ICA)
ACR accreditation process
part 1 – entry application
• qualifications & responsibilities
part 2 - full application
• clinical Images
• phantom Images
continuing requirements of accreditation
phantom & phantom tests
additional annual survey/evaluation tests
ACR accreditation - initial questions????
MARP, Inc.
when is the due date?
has the initial fee been sent?
is there an ACR phantom?
has it been ordered
are there manufacturer
phantoms?
is there a laser printer?
is there a densitometer?
can CD-ROMs be burned?
from scanner?
from PACS?
how much will you do?
evaluation
phantom
QC
paperwork for submission
clinical images
part 1 – entry application
MARP, Inc.
information
• magnet
• practice
credentials
• physicians
• technologists
• physicists
fees
• Accreditation / Reinstate
• $2400 for 1st magnet
• $2300 for subsequent magnets
• Repeat
• $800 for phantom or clinical
• $1600 for both
all magnets at site must be
accredited
45 day submission period begins
qualifications & responsibilities
MARP, Inc.
physician
shall have the responsibility for all aspects of the
study
technologist
qualified medical physicist / MR scientist
should have responsibility for overseeing QC and
monitoring performance
physicist - qualifications
MARP, Inc.
should be certified in appropriate subfield
• recommended: ABR Diagnostic Radiological Physics
• ABMP has certification in MRI
should have responsibility for overseeing QC and
monitoring performance
• physicist/MR scientist not required but strongly recommended
• if not qualified medical physicist/MR scientist → supervising
physician, technologist, or service engineer
MR scientist
• graduate degree in a physical science involving nuclear MR or MRI
• 3 years documented experience in a clinical MRI environment
CME
• 15 hours CME in MRI over previous 3 years
• half must be category I
part 2- full application
MARP, Inc.
ACR clinical image and phantom review
sent by ACR after receiving initial application
• arrives ~ 30 days after initial application
includes MRAP number and due date
• due 45 days after initial application
includes instructions, parameter data forms, and
labels
• everything (but labels) is available on-line
• www.acr.org
reaccreditation application similar
clinical images
MARP, Inc.
required from every magnet at practice location
4 sets of images
original 14 x 17 films
or refilmed from original tapes or discs
electronic submission
• CD-ROM with embedded viewer
• must include functions of:
• window/level
• magnification
• region of interest (including measurement of area, pixel mean, pixel
standard deviation)
• distance measurement
must be obtained within one week (before or after) of the
acquisition of phantom images
different images for whole body vs. cardiac accreditation
clinical images - evaluation
MARP, Inc.
radiologist reviews for:
pulse sequence and image contrast
filming technique
anatomic coverage & imaging planes
spatial resolution
artifacts
exam ID
should be “best work”
pathology is “OK”
volunteers not “OK”
should be complete exam
contrast portion not necessary
clinical images
MARP, Inc.
routine brain examination (for headache)
• sagittal short TR/short TE with dark CSF
• axial or coronal long TR/short TE and long TR/long TE (e.g., long TR double
echo)
routine cervical spine (for radiculopathy)
• sagittal short TR/short TE with dark CSF
• sagittal long TR/long TE or T2*W with bright CSF
• axial long TR/long TE or T2*W with bright CSF
routine lumbar spine ( for back pain)
• sagittal short TR/short TE with dark CSF
• sagittal long TR/long TE or T2*W with bright CSF
• axial short TR/short TE with dark and/or long TR/long TE with bright CSF
routine knee examination (for internal derangement)
• sagittal and coronal with at least one sequence of bright fluid
Note (for SE):
• T1-weighted = short TR/short TE
• PD-weighted = long TR/short TE
• T2-weighted = long TR/long TE
clinical image recommendations
MARP, Inc.
Sequence
Slice thickness
Gap
Maximum Pixel
Dimensions
Brain – Sagittal & Axial
and/or Coronal
≤
5 mm
≤
2 mm
Cervical Spine – Sagittal
≤
3 mm
≤
1 mm
≤
1 mm
Cervical Spine – Axial
≤
3 mm
≤
1 mm
≤
1 mm
Lumbar Spine – Sagittal
≤
5 mm
≤
1.5 mm
≤
1.5 mm
Lumbar Spine – Axial
≤
4 mm
≤
1 mm
≤
1.5 mm
Knee – Sagittal & Coronal
≤
4 mm
≤
1 mm
≤
.75 mm
≤
1.2 mm
clinical images
MARP, Inc.
all images (clinical & phantom) must be submitted
by the due date on the labels sent by the ACR
each of the 4 image types are labeled and placed in
film jackets with completed parameter data form
•
•
•
•
•
pulse sequence
FOV
matrix
slice thickness
slice gap
electronic submission
• 2 CDROMS with embedded viewer
quantitative phantom testing
MARP, Inc.
MRI accreditation phantom
MARP, Inc.
specific ACR MRI accreditation phantom is used
specific protocols for T1 and T2 are provided in the
site instructions with the full application
• phantom site scanning instructions at www.acr.org
each site is required to submit phantom images
using the ACR protocol
• and phantom images using its own routine T1 and T2
weighted scan protocol for head examinations
same phantom images for whole body & cardiac
accreditation
phantom image submission - films & data required
MARP, Inc.
films
12-on-1 films of all 4 series
data
Dicom PC CD-ROM
•
•
•
•
preferred method
each series in separate directory
make certain data are easily located
make certain images are accessible by Osiris software
• do not use embedded viewer
• do not use compressed images
site archive format (tape or disk) is no longer
acceptable
phantom image data
MARP, Inc.
conversion of media to Dicom PC CD-ROM
provided by some
manufacturers
• Fonar
available on newer
workstations / PACS
environments
• make certain images on CD can be
located & opened by Osiris
provided by DesAcc, Inc.
• HTML format on CD-ROM available
for
• Elscint (5¼")
• GE (most 5¼" & most DAT)
• Hitachi (5¼")
• Philips (5¼" & 12")
• Siemens (5¼" & 12")
• Toshiba (5¼")
• $200-250 + $25 shipping
• ~ 2-3 weeks turnaround
• http://www.desacc.com
• 312 930-5617
phantom image data – CD-ROM
MARP, Inc.
available on newer
workstations / PACS
environments
DICOM only CD
with DICOMDIR
MRAP applications
as of Nov 2007:
8798 units
6052 facilities
label CD
MRAP accredited as
of Nov 2007
4604 units
3858 facilities
phantom data CD-ROM
MARP, Inc.
uncompressed DICOM images
256 x 256 x 2 = 131072 bytes
phantom data CD-ROM
MARP, Inc.
embedded readers often compress images
Osiris 4.19
MARP, Inc.
free DICOM viewer
used by ACR reviewers
download at
http://www.sim.hcuge.ch/osiris/01_Osiris_Presentation_EN.htm
make certain Osiris can open images
Osiris 4.19
MARP, Inc.
check:
navigation through images
window & level work
zoom
ROIs
DICOM headers
make certain Osiris does
not crash with your data
most consoles produce
good CD-ROMs
GE
Siemens
Philips
Toshiba
Fonar ?
Hitachi ???
many PACS systems do
not
continuing requirements of accreditation program
MARP, Inc.
weekly QC
annual tests
ACR has the right to perform random site
and/or film inspections
quality control
MARP, Inc.
effective August 2002:
effective July 2005:
weekly tests (formerly
daily):
for renewal, site must
send:
•
•
•
•
•
•
central frequency
transmitter gain /attenuation
geometric accuracy
spatial resolution
low-contrast detectablilty
image artifact assessment
weekly:
• laser film QC
• visual checklist
annual:
• physicist/MR scientist
performance evaluation & QC
review
• 3 months QC/printer data
• annual survey report
• dated within 12 months
• documentation of corrections
for failures
QC and annual survey are
now required for initial
applications too
ACR MRI phantom
MARP, Inc.
J.M. Specialty Parts
11689-Q Sorrento Valley Road
San Diego, CA 92121
(858) 794-7200
$730
delivery
6-8 weeks without MRAP
number
2-3 weeks with MRAP
number
order without MRAP, then
call to add MRAP
Nose
Chin
performance criteria
MARP, Inc.
Phantom Test Guidance
for the ACR MRI
Accreditation Program
•
•
•
•
•
describes tests
instructions
performance criteria
reasons for failure
available from www.acr.org
site scanning instructions
MARP, Inc.
Site Scanning
Instructions for use of
the MRI Phantom for the
ACR MRI Accreditation
Program
• phantom positioning
• pulse sequences
• film and data
instructions
• sent to site with full
application
• available from ACR
phantom pulse sequences
MARP, Inc.
2 ACR sequences
2 site sequences
T1
T1
T2
T2
identical for all sites
site specific
may be used to
increase SNR in order
to pass low-contrast
tests on low-field units
ACR sequences
MARP, Inc.
1 sagittal localizer slice
• T1: Spin Echo, TR=200 ms, TE=20ms, 25 cm FOV, 256x256,
20mm slice, 1 NEX
2 sets of 11 axial slices
• 25 cm FOV, 256x256, 5 mm slice, 5 mm gap, 1 NEX
• T1: Spin Echo, TR=500 ms, TE=20 ms
• T2: Spin Echo, TR=2000 ms, TE=20/80 ms
• Use 2nd echo only
all sites must use the same ACR sequences
• slight modifications allowed if scanner cannot set them
• T1: Spin Echo, TR=515 ms, TE=20 ms (Toshiba Opart)
• document other scan parameters on site data sheet
required phantom tests
MARP, Inc.
geometric accuracy
high-contrast spatial resolution
slice thickness accuracy
slice position accuracy
image intensity uniformity
percent signal ghosting
low-contrast object detectability
image artifacts
alignment of ACR phantom
MARP, Inc.
pre-scanning procedures
MARP, Inc.
magnet should be
checked by service
engineer prior to
acquisition
alignment is important
• center of head coil
•
computer paper
• straight
•
use bubble level
• centered SI, LR & AP
• make localizer slice in all 3
planes
• use grid to check
centering
• record position for future
use
alignment of ACR phantom
MARP, Inc.
sagittal localizer slice
MARP, Inc.
S
P
Nose
1
Nose
I
Chin
Chin
Table
A
Table
11
sagittal localizer slice
MARP, Inc.
set up 11 axial slices for all 4 series
• 5 mm thick with 5 mm gap
notes
• if geometric accuracy is off, low contrast slices 8-11 may
not be accurate
• some units have protocols which number slices opposite of
the ACR recommendations
• may not start renumbering at "1" for each series
• make note of which image is 2nd echo on T2
11 axial slices & localizer
MARP, Inc.
geometric accuracy
MARP, Inc.
sagittal localizer & ACR axial T1 slices 1 & 5
specific window & level:
window as narrow as possible
set level where ½ of water is dark
(mean)
set window width = mean value
& window level = ½ mean value
window/level must be set separately for localizer & axial T1
• both axial slices use same window/level
geometric accuracy - measurements
MARP, Inc.
sagittal localizer
• top-bottom (z)
• 148 mm
action limits
• ± 2 mm
slice 1 & 5 of ACR axial T1
•
•
•
•
horizontal & vertical (x & y)
diagonal on slice 5
190 mm
different W/L than localizer
action limits
• ± 2 mm
geometric accuracy - notes
MARP, Inc.
image may be bowed
• accurate at one location,
inaccurate at another
• measure in the center
bubble may obscure top
of slice
• measure at angle
localizer & axial slices
need different W/L
operator may know the
actual values and aim
for them
open short bore
magnets may use
geometric corrections
• Siemens – Large FOV or
2D distortion correction
high-contrast spatial resolution
MARP, Inc.
1.1 mm
UL
1.0 mm
0.9 mm
LR
measurements
• use slice 1 of ACR T1 & ACR T2
• magnify slice 1 by 2 to 4
• observe UL holes ; adjust window/level
• observe rows: if all 4 holes in a single row are distinguishable, score
image as resolved at this hole size
• view all three sets (1.1 mm, 1.0 mm, 0.9 mm)
• score = smallest holes resolved
• repeat for LR array with columns of holes
performance criteria: 1.0 mm
slice thickness accuracy
MARP, Inc.
slice 1 of ACR T1 & ACR T2
crossed ramps (10:1 slope)
measure mean
• magnify by 2 to 4
• adjust window/level to see signal ramps
• 2 ROIs
• mean of middle of each signal ramp
• take average
measure width
• lower level to ½ average
• set window at minimum
• measure lengths of top and bottom ramps
calculate slice thickness
performance criteria:
• 5.0 ± 0.7 mm
(top × bottom)
slice thickness =0.2×
(top + bottom)
slice thickness - notes
MARP, Inc.
• edges of ramps difficult to determine
• Gibbs artifacts
• noise (low field magnets)
• display may give max-min signal (or use relative scale)
• Hitachi & Toshiba
• use Osiris
• 1 mm measurement error = 1/10 mm error in slice thickness
1.5 T
0.2 T
slice thicknesses measurements
MARP, Inc.
Hitachi Airis II
ROI scale different than
WL scale
“Jump level to ROI mean”
• sets level to mean of ROI
• displays mean using WL
scale
• manually reduce level to ½
• manually reduce width to 1
slice position accuracy
MARP, Inc.
measurements
• use slices 1 & 11 of ACR T1 & ACR T2.
• magnify by 2 to 4 & adjust window/level
• measure difference of left & right bars
• if left bar is longer assign a minus sign to the length.
S
11
11
A
P
1
I
slice position accuracy
MARP, Inc.
performance criteria:
magnitude of bar length
difference ≤ 5 mm.
• actual displacement is 1/2
of the measured difference
(wedges have 45° slopes)
• operator may strive for
more precision than is
necessary
11
image intensity uniformity
MARP, Inc.
slice 7 of ACR T1 & T2
make large ROI (195-205 cm²)
low-signal region:
• set window width to minimum
• lower level until entire ROI is white
• raise level until 1 cm² region of
black appears
• use 1 cm² ROI to record mean of
this low-signal region
high-signal region:
• raise level until only 1 cm² region
of white remains
• use 1 cm² ROI to record mean of
this high-signal region
percent integral uniformity = 100 × 1 −
(high − low)
( high + low)
image intensity uniformity
MARP, Inc.
performance criteria:
PIU ≥ 87.5%
• if there is not a well-defined
high/low intensity level…
…..uniformity is very high!
for 3.0T:
PIU ≥ 82% (July 2005)
image intensity uniformity – 8-channel coils
MARP, Inc.
smaller coil
harder to setup
poorer uniformity
corrections
Surface Coil Intensity Correction
(SCIC) – GE
no correction
SCIC
Prescan Normalization
(Siemens)
“CLEAR”
(Philips)
“Quadrature” vs “SENSE”
percent signal ghosting
MARP, Inc.
use slice 7 of ACR T1
make large ROI (195-205 cm²)*
• record mean
make 4 elliptical ROIs
• 10 cm² with 4:1 ratio
• left, right, top, bottom
• record mean of each
ghosting ratio =
(top + bottom) − (left + right )
(2 × (large ROI)
performance criteria:
• ghosting ratio ≤ 0.025 (2.5%)
low-contrast object detectability
MARP, Inc.
slices 8-11
both ACR series
both site series
adjust window/level for optimum contrast.
1.5T
slice 8
slice 9
slice 10
slice 11
0.35T
low-contrast object detectability
MARP, Inc.
4 slices with low-contrast holes
slices 8-11
• decreasing contrast levels (11→
→ 8)
10 spokes per slice
• 3 holes per spoke
• decreasing size (clockwise)
count complete spokes
all 3 disks must be discernible
• more apparent than background
end with last complete spoke
for phantom review:
• all slices are counted
• total score = discernible spokes
from all four slices
low-contrast object detectability
MARP, Inc.
performance criteria
each ACR series should have a total score of at
least 9 spokes.
for 3.0T, the total score must be at least 37 spokes.
must pass both ACR series or both site series
low-contrast object detectability
MARP, Inc.
causes of failure:
incorrectly positioned
slices
• contrast based on partial
volume averaging
tilted phantom
warped slices / B0
inhomogeneity
correction software
incorrect slice
thickness
ghosting
inadequate SNR
low-contrast object detectability
MARP, Inc.
notes:
several lower spokes may be visible but cannot be
counted due to artifact obscuring one of the higherlevel spokes
window/level each slice separately
start with highest contrast and move down.
site may need to change their protocol (especially
for low field magnets)
low-contrast: high vs. low field
MARP, Inc.
• slice 11 - ACR T1 series
1.5 T
0.3 T
homogeneity
MARP, Inc.
required for annual
survey & initial
evaluation
concerns
geometric distortion?
problems with fat
saturation?
methods
spectral
phase-difference
visual distortion
homogeneity – spectral analysis
MARP, Inc.
measures frequency of
signal-producing
phantom in magnet
DSV
• ↑ homogeneity at center
& with ↓ FOV
•
FOV… phantom size
spectrum displayed on
screen (Hz)
homogeneity
measured in ppm
compare to specs or
baseline values
homogeneity – spectral analysis
MARP, Inc.
manual prescan
GE
EPI tuning
Elscint
FWHM (Hz)
Hz FWHM
ppm =
MHz LarmorFrequency
volume ~ phantom
homogeneity – spectral analysis
MARP, Inc.
Siemens
“system”
“adjustments”
check “confirm
adjustments”
phase-difference (Philips)
MARP, Inc.
shim_check
option under phantom
studies
FFE, fixed technique &
FOV
40 cm cylindrical
phantom
holder to center
phantom
3 orientations
view real images
count B→
→W transitions
shim_check (Philips)
MARP, Inc.
example:
1.5T, FFE, TR=400, TE=16,
30º, 450 mm FOV
B→
→W = 1 ppm
example:
coin taped to phantom
shim_check (Philips)
MARP, Inc.
example:
1.5T, FFE, TR=400, TE=16, 30º, 450 mm FOV
B→
→W = 1 ppm
DHL
homogeneity – geometric distortion
MARP, Inc.
most low-field open magnets
do not have software to check
homogeneity
observe effects on geometric
distortion
• lower bandwidth
warped images
7.4 kHz
3.6 kHz
different bandwidths
MARP, Inc.
measure distortion in frequencyencoding direction with different
bandwidths (Clarke & Chen)
MFH ( ppm) =
(BW1 × BW2 )× (x1 − x2 )
(γ 2π )B0 FOV (BW2 − BW1 )
BW=33 Hz (8448 kHz)
BW=244 Hz (62464 kHz)
FE45 TR=256, TE=45, 7 mm, FA=70°, FOV=20cm
FE5.0 TR=256, TE=5, 7 mm, FA=20°, FOV=20cm
151.6 cm
148.1 cm
SI
2.68 ppm
AP
0.84 ppm
LR
0.31 ppm
SWOMRI
coils
MARP, Inc.
must be checked
annually
SNR
ghosting
uniformity
volume coils
MARP, Inc.
similar to head coil
uniformity
• mean
• high
• low
SNR
• mean
• background noise
ghosting
• mean
• background signal
• ghost signal (PE direction)
phased-array coils
• may be treated as volume
if they have volume
configuration
surface coils
MARP, Inc.
max SNR
• max ROI
• background noise
uniformity
• subjective
ghosting
• subjective
phased-array coils
• may be treated multiple surface
coils if you can distinguish the
location of the arrays
phased-array coils
MARP, Inc.
multiple types
~ volume coil
~ multiple surface coils
(arrays distinguishable)
more complicated
recommendations
MARP, Inc.
be familiar with process
•
•
•
•
read QC manual, phantom guidance & site instructions
set up worksheet or computer program
let site know if they will pass phantom portion
be prepared for more than just phantom tests
scan early in the process
• site may need to adjust protocols prior to acquisition of
clinical and phantom images
• before application to let site know condition of magnet prior
to due date
• magnet may be “in specs” and still fail part of the test
• magnet may need corrections prior to phantom submission
recommendations
MARP, Inc.
physician
• two-week window to get clinical images
• may need time to adopt ACR protocols
service engineer
• magnet should be in top shape prior to obtaining images
• on hand to assist with display and measurement protocols
technologist
• needs assistance in setting up QC
• schedule ~ 5-7 hours
conclusion
MARP, Inc.
physicist
• there is no reason a site should fail the phantom portion of
the MRI Accreditation program if they are being assisted by
a physicist
• phantom materials submitted by physicist should be usable
by ACR
• physicist should be able to perform all tests required by
ACR
MARP, Inc.
Physicist Role in ACR MRI Accreditation
May 2008
Carl R. Keener, Ph.D., DABMP, DABR
keener@MARPinc.com
MARP
Medical & Radiation Physics, Inc.
Department of Radiology
University of Texas Health Science Center at San Antonio